amalgame/commit

third attempt ...

authorJacco van Ossenbruggen
Mon May 12 12:13:33 2014 +0200
committerJacco van Ossenbruggen
Mon May 12 12:13:33 2014 +0200
commit04710dcb95f73c829a233cd409d351f828ed9a68
tree8a5965e34a1f861d1cc13249367e871f7b9053da
parentdac7331068c5630a17279fa65a099049c752459a
Diff style: patch stat
diff --git a/README.md b/README.md
index 40920a1..470b101 100644
--- a/README.md
+++ b/README.md
@@ -1,6 +1,6 @@
 # amalgame
 This is the AMsterdam ALignment GenerAtion MEtatool (amalgame).
-This open source tool has originally been as part of the EuropeanaConnect and PrestoPrime projects at VU University Amsterdam.
+This open source tool has originally been developed as part of the EuropeanaConnect and PrestoPrime projects at VU University Amsterdam.
 
 
 ## Objective
@@ -8,15 +8,15 @@ amalgame provides a web-based interactive platform for creating, analyzing and e
 
 ## amalgame-specific terminology:
 
-- ** Correspondence ** : We define a correspondence as a relationship between two concepts.  In amalgame, the key associated data-structure is the align/3 term, that encodes a series of claims (3rd argument) about the relationship between a source (1st argument) and target concept (2nd argument).  The claims can be objective observations (such as the observation that the prefLabels of both concepts are the same) or interpretations (such as the claim that both concepts refer to the same entity or that one has a skos:broadMatch relationship to the other). Note that align/3 always denotes a 1-1 relationship, you need multiple align/3 terms to represent N-M relationships.  In PROV terms, a correspondence is too fine grained to play a direct role in a PROV graph.  However, each correspondence is part of a mapping dataset, which is a PROV Entity with associated provenance information.  For each claim about two concepts, the align/3 term may record the evidence that was used to make or support that claim.
+- **Correspondence** : We define a correspondence as a relationship between two concepts.  In amalgame, the key associated data-structure is the align/3 term, that encodes a series of claims (3rd argument) about the relationship between a source (1st argument) and target concept (2nd argument).  The claims can be objective observations (such as the observation that the prefLabels of both concepts are the same) or interpretations (such as the claim that both concepts refer to the same entity or that one has a skos:broadMatch relationship to the other). Note that align/3 always denotes a 1-1 relationship, you need multiple align/3 terms to represent N-M relationships.  In PROV terms, a correspondence is too fine grained to play a direct role in a PROV graph.  However, each correspondence is part of a mapping dataset, which is a PROV Entity with associated provenance information.  For each claim about two concepts, the align/3 term may record the evidence that was used to make or support that claim.
 
-- ** Mapping ** : We define a mapping as a homogeneous set of correspondences. These are homogeneous in that they make similar type of claims about their source and target concept.  Because of its homogeneous nature, we hope that evaluating only a small random subset will give reliable insights into the quality of the entire mapping. In amalgame mappings are typically represented as a list of align/3 terms.  This list can be materialized as EDOAL triples in a single named graph.  In PROV terms such a named graph is a single Entity, and all correspondences it contains have the same provenance, e.g. they have all been generated by the same sequence of processing steps.  
+- **Mapping** : We define a mapping as a homogeneous set of correspondences. These are homogeneous in that they make similar type of claims about their source and target concept.  Because of its homogeneous nature, we hope that evaluating only a small random subset will give reliable insights into the quality of the entire mapping. In amalgame mappings are typically represented as a list of align/3 terms.  This list can be materialized as EDOAL triples in a single named graph.  In PROV terms such a named graph is a single Entity, and all correspondences it contains have the same provenance, e.g. they have all been generated by the same sequence of processing steps.  
 
-- ** Alignment ** : An alignment in amalgame is a (typically) hetereogeneous set of correspondences between two vocabularies.  Typically, an alignment is made by merging all mappings of sufficient quality into a single data set.
+- **Alignment** : An alignment in amalgame is a (typically) hetereogeneous set of correspondences between two vocabularies.  Typically, an alignment is made by merging all mappings of sufficient quality into a single data set.
 
-- ** Alignment strategy ** : An alignment strategy is a recipee that defines how the mappings that constitute the alignment have to be made.  It defines for all mappings the amalgame alignment modules that can create them, and their inputs.  In addition to the mappings actually used in the final alignment (these mappings are typically marked with amalgame:status amalgame:final) strategies typically also define (amalgame:intermediate) mappings that are used as input for processes generating other mappings and mappings that are explicitly not used (e.g. for lack of quality) and have been marked with amalgame:discarded.  A strategy defines a striped dependency graph between mappings (which are subclasses of prov:Entities) and amalgame modules (which are subclasses of prov:Activities).  A strategy is also a prov:Plan.
+- **Alignment strategy** : An alignment strategy is a recipee that defines how the mappings that constitute the alignment have to be made.  It defines for all mappings the amalgame alignment modules that can create them, and their inputs.  In addition to the mappings actually used in the final alignment (these mappings are typically marked with amalgame:status amalgame:final) strategies typically also define (amalgame:intermediate) mappings that are used as input for processes generating other mappings and mappings that are explicitly not used (e.g. for lack of quality) and have been marked with amalgame:discarded.  A strategy defines a striped dependency graph between mappings (which are subclasses of prov:Entities) and amalgame modules (which are subclasses of prov:Activities).  A strategy is also a prov:Plan.
 
-- ** Provenance graph ** : The dependency graph of the alignment strategy can be seen as the backbone of a PROV provenance graph.  By executing such a strategy and recording the information that is specific to that execution, one can extend this backbone into a complete PROV graph.  Vice versa, if the PROV graph recording such an execution has been extended by amalgame's strategy vocabulary, the dependency graph that forms the backbone of the PROV graph can be re-used as an amalgame alignment strategy for another run. We model a provenance graph as a prov:Bundle.
+- **Provenance graph** : The dependency graph of the alignment strategy can be seen as the backbone of a PROV provenance graph.  By executing such a strategy and recording the information that is specific to that execution, one can extend this backbone into a complete PROV graph.  Vice versa, if the PROV graph recording such an execution has been extended by amalgame's strategy vocabulary, the dependency graph that forms the backbone of the PROV graph can be re-used as an amalgame alignment strategy for another run. We model a provenance graph as a prov:Bundle.
 
 ## AMALGAME motivation.